A new Kickstarter from Charmed Labs (opens in new tab) aims to bring AI to your camera projects using the Raspberry Pi (opens in new tab) 4 and their own AI Camera. Now your nature cam, physics experiments and data science projects can do all of the hard work for you, starting from $229.
Vizy from Charmed Labs uses the power of the Raspberry Pi 4 to categorize images taken by the AI camera. Detecting animals in your garden, the number of cars driving down your street or identifying species of birds is made possible with Vizy. The high resolution AI camera, which uses the same 12MP Sony sensor as the Official Raspberry Pi HQ Camera, can record at up to 300 frames per second and works even in low light environments. Inside Vizy there is a custom add on board, which provides a screw terminal I/O with support for high current output (up to 1A per channel). The I/O supports digital, analog, PWM and Serial.
To review, control and configure Vizy a web interface is used, for coders there is also a Python framework that can be used to write your own projects with Vizy.
“It (Python) has seen its popularity grow rapidly in recent years and is the most popular language for AI and machine learning. These are some of the reasons we chose Python for Vizy. Tensorflow, OpenCV, PyTorch and other Python libraries are installed and ready for use as well. - Charmed Labs Kickstarter.
With Vizy and our own projects we can set certain actions to occur if a condition is met. For example if you were to count the number of cars driving down a street, this value could be sent to you via SMS, email or via social media. The raw data can be collected and saved to a cloud service, Google Docs for example for later analysis.
Vizy has already tripled its initial funding target and you can also pledge your support from $229 for a Vizy powered by a 2GB Raspberry Pi 4. Extra features such as a zoom lens, outdoor weatherproof enclosures and adapters to use Vizy with telescopes for astronomical projects will add to this price. More details on their Kickstarter page.